Georg Walther
Data Science @E.ON
Signup · Get unlimited contacts
WORK HISTORY
Data Science @E.ON
Berlin, DE
Grid-Scale BESS & Flexibility Market Channel Optimization: Architecting the algorithmic brain of a Virtual Power Plant (VPP) to maximize revenue across heterogeneous assets- Holistic Optimization: Designing Mixed-Integer Linear Programming (MILP) engines that concurrently optimize participation in Day-Ahead, Intraday, and Frequency Response (FCR/aFRR/dynamic services/balancing mechanism) markets- Constraint Modeling: Translating complex physical boundaries (battery degradation, inverter limits) and regulatory codes into hard algorithmic constraints to ensure safe, compliant dispatch- Closed-Loop Execution: Building low-latency feedback loops that connect market forecasts directly to physical asset steering, enabling real-time adaptability- Scalability: developed adaptable logic for diverse asset classes (PV, BESS, EVs) within a unified pooled flexibility model.Hyper-Local Real-Time Carbon Intensity Engine: Engineered a novel grid-intensity estimation system to overcome the inaccuracy of national, post-ex carbon averages- Granular Estimation: Developed an algorithm that computes live, hyper-local CO2 intensity by mapping fine-grained local asset registries against real-time weather-dependent generation (Wind/Solar)- Consumer Steering: Empowering end-users to shift consumption to \"greener\" hours via real-time signals- Production Deployment: The engine currently powers the ÖkoHeld app (Bayernwerk), enabling actionable sustainability for retail consumers: & frameworks: Python, PyDantic, FastAPI, pandas, XGBoost, PyTorch, scikit-learn, mlflow, PuLP, or-tools, linear programming, SCIPAzure Cloud: Azure Function, Azure Data Factory, Azure Machine Learning, Azure Blob Storage, Cosmos DB, Databricks, Delta Table
EDUCATION
ETH Zürich
Bachelor of Science (BSc), Biochemistry
John Innes Centre
Doctor of Philosophy (Ph.D.), Computational Biology
ETH Zürich
Master of Science (MSc), Computational Biology
Science 2 Data Science
Data Science Fellowship
SKILLS
ABOUT GEORG WALTHER
I sit at the intersection of electricity markets, grid physics, and production-grade software.My core expertise is architecting the \"algorithmic brain\" for the modern grid - building automated trading engines and optimization layers that manage heterogeneous assets (BESS, PV, EVs, flexibility pools) across European spot and balancing markets. I move beyond simple \"predictive modeling\" to build deterministic, liability-aware systems that translate complex regulatory constraints into profitable, automated dispatch instructions.Currently focused on the deep digitization of the energy transition- Virtual power plant (VPP) architecture: Designing MILP-based optimization engines that concurrently trade assets across day-ahead, intraday, and FCR/aFRR markets- Asset-aware trading: bridging the gap between financial incentives and physical reality (battery degradation, power and state-of-charge limits, and local grid constraints)- Sustainability intelligence: Engineered hyper-local, real-time CO2 intensity engines to enable demand-side response and greener consumption steering (deployed in the ÖkoHeld app).I don\'t just write code; I build and scale high-performance engineering units- Strategic growth: Built and co-led a profitable data science consultancy unit, growing the team from ground zero to a mix of juniors and PhD specialists- Commercial bridge: Experienced in technical sales, contract negotiation, and translating complex engineering requirements and capabilities into clear business value for C-level stakeholders- Mentorship: Deep experience coaching engineers on architectural patterns, career growth, and navigating the gap between academic theory and production reality.Prior to focusing on energy, I honed my skills on high-frequency, high-dimensional data problems across diverse industries- High-scale time series: Developed anomaly detection and forecasting systems for massive-scale streaming sensor data- Causal inference: Applied propensity scoring and clustering to estimate treatment effects in high-dimensional datasets- Complex event prediction: Built deep learning models for rare-event prediction in user-level time series.Some of the technologies I use- Energy & Optimization: Mixed-Integer Linear Programming (MILP), PuLP, SCIP, Google OR-Tools- Core Engineering: Python (FastAPI, PyDantic, Pandas), Microservices Design, CI/CD- Machine Learning: PyTorch, XGBoost, Scikit-Learn, MLflow, Time-Series Forecasting- Cloud & DevOps: Azure, AWS, Kubernetes, Terraform, Docker.
This profile is compiled from publicly available professional sources. Unifers is not affiliated with or endorsed by LinkedIn. Request removal of this profile.